A dynamic building method of mobile agent migrating path based on Q-learning

Mobile agent is viewed as the proxy of a social member. It is able to carry tasks on the behalf of user and to migrate among hosts in a network and use local service to complete its tasks. The migrating path of mobile agent is a sequence of hosts that the agent migrates along in order to achieve the tasks it carries in network environment. The problem of migrating path planning with multi-objective is a tougher issue. This paper presents a dynamic building method of migrating path based on Q-learning algorithm in large-scale and dynamic open network which is modeled as social acquaintance network. The experimental results show that mobile agent is able to dynamically construct an efficient migrating path by utilizing the Q-learning algorithm step by step.